Receiver for a communication system

Abstract

The present subject matter relates to a receiver including a detector for receiving a signal from a transmitter. The detector includes a set of one or more settable parameters, and circuitry configured for implementing an algorithm having trainable parameters. The algorithm is configured to receive as input information indicative of a status of a communication channel between the transmitter and the receiver and to output values of the set of settable parameters of the detector. The detector is configured to receive a signal corresponding to a message sent by the transmitter and to provide an output indicative of the message based on the received signal and the output values of the set of settable parameters of the detector.

The invention claimed is: 
     
       1. A receiver comprising:
 a detector configured for receiving a signal from at least one transmitter, the detector comprising a set of one or more settable parameters; and 
 circuitry comprising at least one processor and at least one memory including computer program code provided with an algorithm having trainable parameters, the algorithm being configured to receive as input information indicative of a status of a communication channel between the at least one transmitter and the receiver and to output values of the set of settable parameters of the detector, the detector being configured to receive a signal corresponding to a message sent with the at least one transmitter and to provide an output indicative of the message based on the received signal and the output values of the set of settable parameters of the detector; 
 wherein the set of settable parameters of the detector is divided into a number of vectors of a first dimension, wherein the algorithm having trainable parameters is configured to output an equal number of embedding vectors of a second dimension, the second dimension being smaller than the first dimension, wherein the algorithm is configured to generate the values for the settable parameters from the embedding vectors using a trainable matrix having a dimension comprising the first dimension and the second dimension. 
 
     
     
       2. The receiver of  claim 1 , wherein the detector is configured to implement an iterative message recovery algorithm or a neural network. 
     
     
       3. The receiver of  claim 2 , wherein the values of the set of one or more settable parameters are shared between iterations of the iterative message recovery algorithm. 
     
     
       4. The receiver of  claim 2 , the iterative message recovery algorithm being any one of: an orthogonal approximate message passing network or a deep learning multiple-input and multiple-output detection scheme. 
     
     
       5. The receiver of  claim 1 , the algorithm having trainable parameters being a trained machine learning model. 
     
     
       6. The receiver of  claim 1 , the receiver comprising multiple receiving antennas, wherein the information indicative of the status of the communication channel is an estimate of a multiple-input and multiple-output channel matrix. 
     
     
       7. The receiver of  claim 6 , the detector comprising a demodulator and a demapper, the demodulator being configured to provide a probability that a modulated symbol of the message is transmitted from a particular transmitting antenna of multiple transmitting antennas given the received signal; the demapper being configured to provide a probability that each bit of the modulated symbol is set to one with a particular transmitting antenna of the multiple transmitting antennas, wherein the provided output comprises any one anyone of the two probabilities. 
     
     
       8. The receiver of  claim 1 , the algorithm having trainable parameters being a neural network. 
     
     
       9. A method comprising:
 providing an algorithm having trainable parameters that is configured to receive as input information indicative of a status of a communication channel between at least one transmitter and a receiver and to output values of a set of one or more settable parameters of a detector, 
 receiving a signal at the receiver, 
 in response to inputting the information to the algorithm, receiving from the algorithm values of the set of settable parameters, and 
 inputting the received signal and the values of the set of settable parameters to the detector for receiving an output indicative of a message transmitted to the receiver; 
 wherein the set of settable parameters of the detector is divided into a number of vectors of a first dimension, wherein the algorithm having trainable parameters is configured to output an equal number of embedding vectors of a second dimension, the second dimension being smaller than the first dimension, wherein the algorithm is configured to generate the values for the settable parameters from the embedding vectors using a trainable matrix having a dimension comprising the first dimension and the second dimension. 
 
     
     
       10. The method of  claim 9 , further comprising
 receiving a training set comprising triplet samples, each triplet sample comprising a channel state information, a transmitted message and a received signal; 
 training the algorithm using the training set and the detector, thereby generating the algorithm with values of the trainable parameters. 
 
     
     
       11. The method of  claim 10 , the at least one transmitter comprising multiple transmitting antennas and the receiver comprising multiple receiving antennas, the message comprising multiple bit vectors each being transmitted with a respective transmitting antenna of the multiple transmitting antennas, the training being performed to obtain values of the trainable parameters with optimizing a predefined loss function, the loss function being a binary cross entropy on transmitted bits or a categorical cross entropy on symbols. 
     
     
       12. The method of  claim 11 , the training further providing values of trainable matrices representing constellations of the at least one transmitter, thereby enabling a joint optimization of the receiver and the at least one transmitter. 
     
     
       13. The method of  claim 9 , the detector is configured to implement an iterative message recovery algorithm, the values of the set of settable parameters being shared between iterations of the algorithm. 
     
     
       14. A computer program product comprising a non-transitory memory encoded with instructions stored thereon configured for causing an apparatus to perform at least the following:
 providing an algorithm having trainable parameters that is configured to receive as input information indicative of a status of a communication channel between at least one transmitter and the apparatus and to output values of a set of one or more settable parameters of a detector of the apparatus, 
 receiving a signal at the apparatus, 
 in response to inputting the information to the algorithm, receiving from the algorithm values of the set of settable parameters, and 
 inputting the received signal and the values of the set of settable parameters to the detector for receiving an output indicative of a message transmitted to the apparatus; 
 wherein the set of settable parameters of the detector is divided into a number of vectors of a first dimension, wherein the algorithm having trainable parameters is configured to output an equal number of embedding vectors of a second dimension, the second dimension being smaller than the first dimension, wherein the algorithm is configured to generate the values for the settable parameters from the embedding vectors using a trainable matrix having a dimension comprising the first dimension and the second dimension. 
 
     
     
       15. An apparatus comprising a receiver and at least one transmitter, the receiver comprising:
 a detector configured for receiving a signal from the at least one transmitter, the detector comprising a set of one or more settable parameters, and 
 circuitry comprising at least one processor and at least one memory including an algorithm having trainable parameters, the algorithm being configured to receive as input information indicative of a status of a communication channel between the at least one transmitter and the receiver and to output values of the set of settable parameters of the detector, the detector being configured to receive a signal corresponding to a message sent with the at least one transmitter and to provide an output indicative of the message based on the received signal and the output values of the set of settable parameters of the detector; 
 wherein the set of settable parameters of the detector is divided into a number of vectors of a first dimension, wherein the algorithm having trainable parameters is configured to output an equal number of embedding vectors of a second dimension, the second dimension being smaller than the first dimension, wherein the algorithm is configured to generate the values for the settable parameters from the embedding vectors using a trainable matrix having a dimension comprising the first dimension and the second dimension. 
 
     
     
       16. The apparatus of  claim 15 , the receiver comprising multiple receiving antennas, the at least one transmitter being one of: one transmitter having multiple transmitting antennas or multiple transmitters each having a transmitting antenna, wherein the information indicative of the status of the communication channel is a multiple-input and multiple-output channel matrix. 
     
     
       17. A receiver comprising:
 a detector configured for receiving a signal from at least one transmitter, the detector comprising a set of one or more settable parameters; and 
 circuitry comprising at least one processor and at least one memory including computer program code provided with an algorithm having trainable parameters, the algorithm being configured to receive as input information indicative of a status of a communication channel between the at least one transmitter and the receiver and to output values of the set of settable parameters of the detector, the detector being configured to receive a signal corresponding to a message sent with the at least one transmitter and to provide an output indicative of the message based on the received signal and the output values of the set of settable parameters of the detector; 
 wherein the receiver comprises multiple receiving antennas, wherein the information indicative of the status of the communication channel is an estimate of a multiple-input and multiple-output channel matrix; 
 wherein the detector comprises a demodulator and a demapper, the demodulator being configured to provide a probability that a modulated symbol of the message is transmitted from a particular transmitting antenna of multiple transmitting antennas given the received signal, and the demapper being configured to provide a probability that each bit of the modulated symbol is set to one with a particular transmitting antenna of the multiple transmitting antennas, wherein the provided output comprises any one of the two probabilities. 
 
     
     
       18. An apparatus comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to:
 provide an algorithm having trainable parameters that is configured to receive as input information indicative of a status of a communication channel between at least one transmitter and the apparatus and to output values of a set of one or more settable parameters of a detector of the apparatus, 
 receive a signal at the apparatus, 
 in response to inputting the information to the algorithm, receive from the algorithm values of the set of settable parameters, and 
 input the received signal and the values of the set of settable parameters to the detector for receiving an output indicative of a message transmitted to the apparatus; 
 wherein the apparatus comprises multiple receiving antennas, wherein the information indicative of the status of the communication channel is an estimate of a multiple-input and multiple-output channel matrix; 
 wherein the detector comprises a demodulator and a demapper, the demodulator being configured to provide a probability that a modulated symbol of the message is transmitted from a particular transmitting antenna of multiple transmitting antennas given the received signal, and the demapper being configured to provide a probability that each bit of the modulated symbol is set to one with a particular transmitting antenna of the multiple transmitting antennas, wherein the provided output comprises any one of the two probabilities.